Prevention and treatment of viral infections is a major medical goal. Diseases caused by non-passageable (NP) viruses are particularly intractable because the causative agents cannot be readily grown in laboratories for investigation of disease-causing mechanisms or for development of vaccines or therapies. The diseases caused by NP viruses are often quite serious, including cancers, derangement of the immune system, and debilitation.
Hepatitis B virus, or HBV, is a highly infectious NP virus of humans. In many African and Asian countries, HBV infection is endemic. Large portions of these populations are carriers of the virus and, therefore, spread of the disease is extremely difficult to control. In more developed countries, blood transfusion is one of the major modes of transmission.
HBV contains three surface glycoproteins of different size that are encoded by a single open reading frame. Each utilizes a different ATG start codon but all three employ the same TAA stop signal. Ganem & Varmus, 1987; Robinson, 1990. For HBV strain ayw, the small surface antigen, S-sAg, or major protein, contains 226 amino acids. The middle surface glycoprotein, preS2-S-sAg or middle protein has an additional 55 amino acids at the amino terminus of the S-sAg. Likewise, the large surface glycoprotein, preS1-S2-S-sAg or large protein, has 108 amino acids linked to the amino terminus of the preS2-S-sAg. The use of the terms "preS1" or "preS2" regions hereafter are references to the additional 108 and 55 amino acid segments seen in the large and middle antigens, respectively.
The large protein is myristylated at its N-terminal glycine residue. Persing et al. (1987). The molecular weight of each antigen depends on glycosylation and is 24 or 27 kD for the small protein, 30 or 33 kD for the middle protein and 39 or 42 kD for the large protein. Ganem & Varmus, 1987.
The preS regions bind to receptors on the surface of liver cells either via polymerized human serum albumin, in the preS2 region, or by direct binding of the preS1 region to the receptor molecule. Pontisso et al., 1989a; Neurath et al., 1990; Pontisso et al., 1989b; Petit et al., 1992. PreS1-S2-S-sAg remains associated with the endoplasmic reticulum (ER) and leaves the infected cell only as a component of an infectious virus particle (22 nm) or as a component of filamentous, non-infectious particles (42nm) late in infection. Hollinger, 1990. By contrast, the middle and the major antigen are secreted efficiently and assemble into infectious 22 nm particles. Mechanisms that mediate translocation of the middle and the major antigen through the membrane of the ER, or that prevent secretion of the large antigen, have been described. Ou & Rutter, 1987; Masuda et al., 1990; Prange et al., 1992.
All three HBV surface proteins induce B cell- and T cell-specific immune response in infected individuals. In approximately 90% of cases, these responses result in a complete clearance of infectious virus. Hollinger, 1990. Detailed analysis of the surface glycoproteins has identified discrete B cell- and T cell-specific neutralizing epitopes in both the major antigen and the preS regions. Neurath et al., 1985; Milich et al., 1985a; Milich et al., 1985b; Neurath et al., 1989.
Hepatitis B virus antigen consists of a group-specific antigen, a, and at least two subtype determinants, d/y and w/r (Le Bouvier, 1971; Bancroft et al., 1972) resulting in several distinct antigenic genomic subtypes: adw, ayw, adr and ayr. The geographic distribution of subtypes has been described by Nishioka et al. (1975). Subtype cross-protection has been shown by Szmuness et al. (1982). Some of the subtype-specific determinants are located in the preS regions (e.g., Milich et al., 1990a & 1990b).
Vaccination studies with several mouse strains have shown that the response to certain S-sAg epitopes is dependent on the presence of defined molecules of the major histocompatibility complex. Certain strains not responding to the major protein, however, do respond to additional epitopes in the preS regions of the large protein. Nonresponsiveness of mice to vaccines based on the major antigen alone, therefore, can be circumvented by including the large protein in vaccine preparations. Milich et al., 1985a & 1985b.
In humans, the immune response varies from nonresponding to low-responding to high-responding. This effect appears to be haplotype-dependent. Egea et al. (1991). While vaccines based on the major antigen alone are, in most cases, sufficient to induce an antibody response in humans resulting in protection from disease (Szmuness et al., 1980), a certain percentage of hosts, do not respond to S-sAg-based vaccines. In one study, 7-15% of those vaccinated with S-sAg did not respond by producing antibodies. Nowicki et al., J. Infect. Dis. 152: 1245-58 (1985). Vaccines partially, primarily or exclusively based on the large antigen are therefore desirable for the vaccination of patients that respond poorly to vaccines containing only the major antigen.
In some reports, combinations of antigens have been found to show improvement in responder rates, suggesting that to be most effective, a candidate vaccine should contain the antigenic components of all three surface glycoproteins. To determine the optimum combinations of antigens and optimum ratios of those antigens, it is necessary to be able to isolate and purify various antigenic components of HBV, and to have sufficient quantities to conduct appropriate research.
Despite intensive efforts in this regard, the antigenic structure and function of various HBV antigens remains largely unknown. One of the barriers to gaining information on HBV antigens is the difficulty of obtaining sufficient quantities these polypeptides. A further problem is the isolation of individual HBV components for the development of vaccines and drugs of general efficacy against a variety of infectious HBV strains.
Some components of the antigens may be painstakingly isolated from serum or tissues of infected humans or other primates, but these efforts yield unacceptably small quantities. Antigens prepared from infected cells, blood or purified virus also run the risk of contamination with active virus. Therefore, it appears that isolation and purification of antigenic components of HBV in useful quantities is feasible only by recombinant technology. Large quantities and more pure samples of antigenic components relative to those obtainable from tissues of infected individuals are expected from recombinant methods. Thus, preparation of vaccines and development of therapies should be aided by the improved production and purification capabilities afforded by in vitro expression of recombinant HBV proteins.
Expression of the small, middle and large HBV surface glycoprotein in various prokaryotic and eukaryotic expression systems has been demonstrated. Yu et al. 1990; Lin et al. 1991; Schodel et al. 1991; Kumar et al. 1992. These systems include yeast (Dehoux et al. 1986; Imamura et al. 1987; Shiosaki et al. 1991; Kuroda et al. 1992), mammalian cells (Ou & Rutter, 1987; Lee et al. 1989; Youn et al. 1989; Korec et al. 1992) and viral expression systems (McLachlan et al. 1987; Yuasa et al. 1991; Shiraki et al. 1991; Belyaev et al. 1991). However, yeast expression systems produce overglycosylated HBV large antigens and non-glycosylated major proteins, neither of which are secreted. Prokaryotic systems do not glycosylate polypeptides nor do they secrete recombinant proteins.
The current system for commercial expression of HBV surface glycoproteins for the production of vaccines involves a yeast expression system that constitutively expresses these proteins. Transformed yeast cells produce the small HBV surface glycoprotein with characteristics similar to the 22 nm particles seen in the blood of naturally-infected persons. The major protein produced in yeast, however, is not glycosylated (see EP 0 414 374 A2, p. 16), while the large antigen is overglycosylated. To overcome this disadvantage, large antigens with deleted glycosylation sites have been constructed (EP 0 414 374 A2, Example F.3). Such modifications, however, are likely to reduce the natural immunogenicity of the antigen.
In natural infections with HBV, the large surface glycoprotein is translated from a preS1-specific mRNA, whereas the middle and the small protein are translated from a preS2-specific mRNA (Robinson 1990). The transcriptional promoter elements for the preS1-S2-S gene are located upstream of this gene while the preS2-S gene has its own promoter elements located within the preS1 region. In transformed cell lines containing an integrated preS1-S2-S gene, both the preS1-S2-S 2.4 kB mRNA and the preS2-s 2.1 kB mRNA are transcribed, and all three antigens are expressed. Permanent cell lines having incorporated the preS1-S2-S gene secrete particles consisting predominantly of the major S antigen with varying amounts of the M and the L antigen (PCT WO 90/10058). HBV surface glycoproteins expressed in eukaryotic cells are secreted more efficiently, but the yield of recombinant protein is less than in prokaryotes or yeasts. Thus far, only Chinese Hamster Ovary (CHO) cells have shown continuous expression of recombinant HBV surface glycoprotein genes. Ou & Rutter, 1987; Lee et al., 1989; Youn et al., 1989; Korec et al., 1992.
In the vaccinia virus expression system, HBV surface glycoproteins have been expressed mainly under the control of the weak vaccinia P7.5 promoter. Since expression levels using this promoter are low, the large antigen has been detected only using sensitive detection methods such as radioactive labeling and autoradiography. Cheng et al., 1986; Cheng et al., 1987; Kutiniva et al., 1990; Nemeckova et al., 1991. It was reported that expression of the small and middle antigens in a vaccinia virus system results in a efficient secretion of both polypeptides. Cheng et al., 1987. The large antigen expressed via vaccinia virus, however, is not secreted, but remains associated with intracellular membranes. Cheng et al., 1986.
In an effort to improve results using vaccinia virus as an expression vehicle, Fuerst et al. (1987) replaced the normal P7.5 promoter with a bacteriophage T7 promoter. In addition, a T7 RNA polymerase gene was incorporated into a separate plasmid or virus to facilitate transcription from the heterologous T7 promoter. The results showed that with three different proteins, E. coli .beta.-galactosidase, HBV small antigen and HIV gp160, expression from a T7 promoter by a T7 polymerase was higher than that seen with native vaccinia promoters.
Elroy-Stein et al. (1989) provided yet another variation on the vaccinia virus expression system. Using the hybrid T7/vaccinia system described above, they further included an encephalomyocarditis virus 5' untranslated region (UTR) to help improve translation of the RNA produced by the T7 polymerase. The results showed improved expression of a recombinant chloramphenicol transferase (cat) gene with protein levels approaching 10% of total cell protein.
An interesting result of cloning of the preS1-preS2-S-sAg gene into a vaccinia expression vector is expression of the large antigen without expression of the small or middle antigens. This is in contrast to transformed cell lines where all three antigens are expressed from the preS1-preS2-S-sAg gene. Thus, the present invention shows that vaccinia systems allow for the expression of pure preparations of the large protein alone, increasing the content of the important preS1-region epitopes.
In contrast to virtually all other reports, Li et al., U.S. Pat. No. 5,077,213, demonstrates that a recombinant vaccinia virus (strain Guang), carrying the preS1-S2-S gene from HBV strain adr, drives the efficient secretion of small, middle and large antigen from TK.sup.- osteosarcoma cells infected with this recombinant. It is not apparent from Li et al. what factors account for the reported results when in other reports, vaccinia is (i) shown not to secrete large antigen and (ii) fails to produce small and middle antigens from the native HBV promoter. In addition, it is difficult to discern any general principle(s) in this regard because Li et al. used neoplastic (osteosarcoma) cells as hosts. The unusual characteristics of these cells may well be responsible for the different results reported by Li et al.